Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Aug 1.
Published in final edited form as: Drug Alcohol Depend. 2020 May 25;213:108081. doi: 10.1016/j.drugalcdep.2020.108081

Inpatient addiction consultation and post-discharge 30-day acute care utilization

Zoe M Weinstein a,b,*, Debbie M Cheng c, Maria J D’Amico a, Leah S Forman d, Danny Regan a,b, Alexandra Yurkovic a,b, Jeffrey H Samet a,b,e, Alexander Y Walley a,b
PMCID: PMC7371521  NIHMSID: NIHMS1599421  PMID: 32485657

Abstract

Background

Addiction Consult Services care for hospitalized patients with substance use disorders (SUD), who frequently utilize costly medical services. This study evaluates whether an addiction consult is associated with 30-day acute care utilization.

Methods

This was a retrospective cohort study of 3905 inpatients with SUD. Acute care utilization was defined as any emergency department visit or re-hospitalization within 30 days of discharge. Inverse probability of treatment weighted generalized estimating equations logistic regression models were used to evaluate the relationship between receipt of an addiction consult and 30-day acute care utilization. Exploratory subgroup analyses were performed to describe whether this association differed by type of SUD and discharge on medication for addiction treatment.

Results

The 30-day acute care utilization rate was 39.5% among patients with a consult and 36.0% among those without. Addiction consults were not significantly associated with care utilization (Adjusted Odds Ratio 1.02; 0.82, 1.28). No significant differences were detected in subgroup analyses; however, the decreased odds among patients with OUD given medication was clinically notable (AOR 0.69; 0.47, 1.02).

Discussion

Repeat acute care utilization is common among hospitalized patients with SUD, particularly those seen by the addiction consult services. While this study did not detect a significant association between addiction consults and 30-day acute care utilization, this relationship merits further evaluation using prospective studies, controlling for key confounders and with a focus on the impact of medications for opioid use disorder.

Keywords: Consult service, Substance use disorder, Care utilization, Hospital readmission, Medication for addiction treatment

1. Introduction

The United States of America is in the midst of an overdose epidemic. (Jones CM et al., 2013; Rudd et al., 2016; Volkow et al., 2014) Despite the increased public attention to substance use disorders (SUD), less than a quarter of the 20 million Americans with SUD received any treatment in 2017.(Substance Abuse and Mental Health Services Administration, 2018) Barriers to SUD care include lack of patient and provider awareness of evidence-based treatment options, lack of integration of addiction care into mainstream medicine and stigma.(Appel et al., 2004; Appel and Oldak, 2007; Drainoni et al., 2014; Noska et al., 2015; Walley et al., 2008; Yarborough et al., 2016) Even hospitalized patients with active co-morbid SUD do not routinely have their SUD addressed by the inpatient care team, contributing to increased costs and readmission risk (Baser et al., 2011; Lynch et al., 2014; Walley et al., 2012). In 2016 there were 657,000 hospital admissions in the USA with a substance use disorder as the primary diagnosis, with a total medical cost of $5.2 billion, which does not account for other societal costs such as lost productivity.(Agency for Healthcare Research and Quality, 2019) SUD are also highly prevalent in hospitals, with nearly one fifth of all hospitalized patients having a substance use disorder as a primary or secondary diagnosis. (Center for Health Information And Analysis, 2016; Walley et al., 2012)

Patients with substance use are more likely to frequently utilize inpatient services (Bell et al., 2017; Robinson et al., 2016; Wu et al., 2018) and have preventable readmissions.(Goldfield et al., 2008; Gupta et al., 2018; McIntyre et al., 2016) Infectious complications from injection drug use are also on the rise, driving increased care utilization among people with SUD.(Gray et al., 2018) Some evidence suggests that addressing substance use during hospitalization can improve care utilization after leaving the hospital.(Baser et al., 2011; Lynch et al., 2014; Smith-Bernardin et al., 2018; Wei et al., 2015)

An Addiction Consult Service (ACS) is a multi-disciplinary consultation team with addiction expertise that takes advantage of the reachable moment of a hospitalization to diagnose patients with SUD, counsel them about treatment options, coach and collaborate with inpatient providers, initiate evidence-based medications for addiction and bridge patients to long-term outpatient treatment.(Priest and McCarty, 2019; Shanahan et al., 2010; Weinstein et al., 2018) This model of care is rapidly expanding, but is still relatively new. Therefore, the effectiveness of these real-world interventions on post-discharge care utilization has not been well described. One recent study showed decreases in self-reported care utilization among patients who had received an addiction consult.(Wakeman et al., 2017) A second study showed decreased 90-day readmissions for a subset of patients with a serious infection secondary to drug use requiring long-term antibiotics. (Marks et al., 2019) However, another study did not show an impact of an ACS on care utilization in a retrospective chart review.(Nordeck et al., 2018)

The primary aim of this study was to evaluate whether an addiction consult is associated with acute care utilization, including emergency department and inpatient admissions, within 30 days of discharge among patients with SUD. An exploratory aim of this study was to evaluate whether the impact of an addiction consult differed by the type of SUD and/or the receipt of medications for addiction treatment.

2. Materials and Methods

2.1. Patients and admissions

The INREACH (INpatient REadmission post-Addiction Consult Help) Study was a retrospective cohort study of all admitted non-pregnant adult patients with a SUD diagnoses from July 2015 to July 2016 at Boston Medical Center, a large urban safety net hospital. Primary and secondary discharge diagnosis codes were used to identify patients with SUD. Types of SUD diagnoses included opioid, cocaine, alcohol, cannabis, sedative and stimulant. The sample included both inpatient admissions and observation-status admissions. A single patient could contribute multiple admissions. For a given admission (either inpatient or observation), a patient was considered to be in the exposed group if he or she received an addiction consult during that admission, as documented by a note from the ACS and confirmed by manual chart review by a trained medical student (M.D.).

2.2. Measures

Basic demographics, medical diagnoses, admission characteristics and administered medications were abstracted from the Electronic Medical Record. Demographics included age, sex, race/ethnicity, insurance type, and housing status. Admission characteristics included the admitting service, length of stay, whether Intensive Care Unit or Psychiatry consult was required and whether the patient left against medical advice (AMA). Medical and substance use characteristics included HIV status, types of SUD and number of SUD, whether naloxone was administered or an overdose occurred, acute complications of substance use (e.g. seizure, endocarditis), whether the patient was receiving buprenorphine or methadone as an outpatient, if the Clinical Opiate Withdrawal Scale (COWS) was used and if the patient received inpatient buprenorphine, methadone, benzodiazepines or phenobarbital. Medications for addiction treatment on discharge were also recorded, including the start or continuation of methadone, buprenorphine, naltrexone, acamprosate, disulfiram or topiramate, as these were the common medications recommended by the ACS.(Trowbridge et al., 2017)

2.2.1. Main independent variable and primary outcome

The main independent variable was whether or not the patient received an addiction consult during the given encounter. The primary outcome was any 30-day post discharge acute care utilization, measured as either an emergency department visit or a new admission (inpatient or observation) within 30 days of discharge from the index admission. This care utilization was not limited to SUD-related admissions.

2.3. Statistical analyses

Descriptive statistics were calculated including frequencies and percentages of categorical variables and means, standard deviations, medians, 25th and 75th percentiles for continuous variables. To address potential confounding, inverse probability of treatment weighted (IPTW) generalized estimating equations (GEE) logistic regression models were used to examine the relationship between receipt of consultation from the ACS and 30-day acute care utilization. The GEE model accounts for the correlation due to including multiple admissions from individual patients. The models were fit using a logit link and an autoregressive working correlation structure. Robust empirical standard errors are reported.

Weights were derived from propensity score modeling of receipt of consultation from the ACS as a function of potential confounders. To estimate the propensity of having an addiction consult in a given admission, a longitudinal GEE regression model was fit using receipt of an addiction consult as the outcome of the model. We included in this model several demographic, admission, substance use and medical characteristics (Table 1) as they were considered potential confounders of the association between acute care utilization and an addiction consult. These variables, listed in the Measures section, were selected a priori based on current literature and guided by clinical experience of the authors.(Chuang et al., 2017; D’Amico et al., 2019; McIntyre et al., 2016; Nordeck et al., 2018; Wei et al., 2015; Wu et al., 2018) Spearman correlation coefficients were calculated between independent variables to assess potential collinearity. No pair of variables included in the models had correlation >0.50.

Table 1.

Characteristics of inpatients with a substance use disorder who did and did not receive an addiction consult used to calculate the propensity of receiving an addiction consult (July 2015 – July 2016), including unweighted and weighted standardized differences

Unweighted sample (and N)* Weighted samplea*
Variable Measure, %(N) Consult
N= 436		 No Consult
N= 3469		 Standardized
Differenceb 		Consult
N= 436		 No Consult
N= 3469		 Standardized
Differenceb
Basic demographics
Age Mean (Std Dev)
Median (25th, 75th)		 43.0 (12.2)
43.0 (32, 53)		 49.6 (14.6)
51.0 (39, 59)		 0.49 47.2 (30.0)
49.0 (38, 56)		 48.9 (15.4)
50.0 (38, 59)		 0.07
Sex Male 68.3% (298) 70.7% (2451) 0.05 67.5% 70.9% 0.07
Female 31.7% (138) 29.3% (1018) 0.05 32.5% 29.1% 0.07
Race/ethnicity White 51.4% (224) 40.1% (1391) 0.23 46.7% 41.8% 0.10
African American 26.1% (114) 38.9% (1351) 0.28 32.3% 37.4% 0.11
Latino 19.3% (84) 16.6% (577) 0.07 17.7% 16.7% 0.03
Other 2.8% (12) 3.6% (126) 0.05 3.0% 3.5% 0.03
Asian 0.5% (2) 0.7% (24) 0.03 0.2% 0.7% 0.06
Insurance status Medicaid 73.2% (319) 55.6% (1928) 0.37 64.6% 57.3% 0.15
Medicare 16.3% (71) 24.9% (863) 0.21 18.7% 24.0% 0.13
Private 8.7% (38) 15.9% (551) 0.22 12.8% 15.3% 0.07
Uninsured 1.8% (8) 3.7% (127) 0.11 3.9% 3.4% 0.02
Homeless 38.1% (166) 22.3% (772) 0.35 29.7% 24.0% 0.13
Admission characteristics
Admitting Service General and Family Medicine 61.7% (269) 45.2% (1567) 0.34 58.6% 46.2% 0.25
Medical Subspecialties 11.2% (49) 10.7% (372) 0.02 10.5% 10.9% 0.01
General Surgery and Surgical Subspecialties 8.9% (39) 22.2% (771) 0.37 14.0% 20.8% 0.18
Intensive Care Unit 16.7% (73) 9.9% (342) 0.20 14.5% 11.2% 0.10
Otherc 1.4% (6) 12.0% (417) 0.44 2.4% 10.9% 0.35
Length of stay Mean (Std Dev)
Median (25th, 75th)		 5.9 (6.9)
3.6 (2, 7)		 4.1 (7.4)
2. (1, 4)		 0.25 5.3 (13.2)
3.5 (2, 7)		 4.5 (9.3)
2.2 (1, 5)		 0.07
ICU stay during admissiond 25.9% (113) 19.6% (681) 0.15 24.6% 20.9% 0.09
Psychiatry consulted during admission 22.7% (99) 11.8% (408) 0.29 16.5% 13.7% 0.08
Discharged against medical advice 11.5% (50) 8.5% (295) 0.10 11.5% 8.9% 0.09
Medical and substance use characteristics
HIV Positive 19.3% (84) 6.5% (226) 0.39 17.1% 7.6% 0.29
Type of substance use disordere Non-Opioid 21.8% (95) 71.8% (2492) 1.16 48.8% 66.9% 0.37
Opioid Only 24.3% (106) 16.0% (555) 0.21 24.7% 16.6% 0.20
Opioid and Other Use Disorder 53.9% (235) 12.2% (422) 0.99 26.6% 16.5% 0.25
Number of
substance use
disorder diagnoses		 Mean (Std Dev)
Median (25th, 75th)		 1.9 (1.0)
2.0 (1, 2)		 1.3 (0.6)
1.0 (1, 1)		 0.78 1.5 (2.0)
1.0 (1, 2)		 1.4 (0.7)
1.0 (1, 2)		 0.11
Naloxone administered 3.2% (14) 1.5% (51) 0.12 3.2% 1.7% 0.09
Overdose diagnosis 10.8% (47) 7.1% 9247) 0.13 6.6% 8.0% 0.05
Acute complicationsg 57.1% (249) 35.3% (1224) 0.45 43.6% 37.9% 0.11
COWS orderede 33.0% (144) 2.9% (102) 0.85 9.8% 5.8% 0.15
Outpatient methadone 3.0% (13) 2.2% (78) 0.05 3.5% 2.4% 0.07
Outpatient buprenorphine 6.2% (27) 2.6% (90) 0.18 4.5% 3.2% 0.07
Methadone ordered during admissione 49.3% (215) 8.6% (300) 1.00 17.9% 12.8% 0.14
Buprenorphine ordered during admissione 11.5% (50) 4.3% (150) 0.27 6.1% 5.4% 0.03
Benzodiazepines/barbiturates ordered during admission 48.4% (211) 35.9% (1244) 0.26 50.8% 37.6% 0.27
Open in a new tab

Legend

*

All analyses performed reflect patient-level analyses

a

weights are truncated at 99th percentile due to large weights

b

The standardized difference was calculated as the difference in means (or proportions) divided by the pooled standard deviation

c

Other services include Neurology, Gynecology, Pediatrics, Unknown

d

Patient stayed at least one day in the Intensive Care Unit during admission

e

Denotes variables removed from propensity score for secondary analysis of effect modification

f

Acute complications include: cellulitis/abscess, osteomyelitis, tenosynovitis, endocarditis, acute and chronic pancreatitis, delirium tremens, acute and chronic pancreatitis, seizures and Wernicke-Korsakoff Syndrome

g

Clinical Opiate Withdrawal Scale ordered during admission

The resulting model calculated the predicted probability of having an addiction consult (i.e., the propensity score for each patient encounter). To assess for balance in covariates, descriptive statistics for each covariate by exposure group (ACS consult yes vs. no) were reviewed for the unweighted and weighted samples. (Table 1) The propensity scores were then incorporated into the analyses using IPTW logistic regression models and robust standard errors were reported. Due to large weights in the data we truncated conventional weights at the 99th percentile.(Austin and Stuart, 2015) We also performed sensitivity analyses including stabilized weights (i.e. conventional weights multiplied by marginal probability of receiving given exposure), both without truncation and truncated at a value of 10.(Robins et al., 2000; Stürmer et al., 2014) We performed an additional sensitivity analysis to explore the impact of variables of unknown temporality included in the propensity score, specifically variables that could have occurred before or as a consequence of an addiction consult (e.g. the ordering of methadone during the admission).

2.3.1. Exploratory analyses

Pre-planned exploratory analyses were performed to describe the association between the receipt of a consult from the ACS and any post-discharge 30-day acute care utilization stratified by type of SUD diagnosis (opioid vs non-opioid). In addition, we hypothesized that a major impact of the addiction consult would be through medications (specifically medications for opioid use disorder). Thus, we conducted an additional analysis stratified by a 4-category effect modifier: 1) Opioid use disorder (OUD) and medication started/continued; 2) OUD and no medication started/continued; 3) non-OUD and medication started/continued; and 4) non-OUD and no medication started/continued. For these analyses we re-ran the propensity score analysis, excluding the OUD and medication variables as these were used for stratification. We performed all analyses using SAS version 9.4 (SAS Institute, Inc., NC, USA). This study was approved by the Boston University Medical Campus Institutional Review Board.

3. Results

There were 5979 total admissions for 3905 unique patients with SUD during the study period. An addiction consult was performed in only 694 (11.6%) of these visits. The demographics of the patients who received a consult, as compared with those who did not, have been previously described.(D’Amico et al., 2019) In brief, the following patient characteristics were associated with higher odds of receiving an addiction consult: opioid use disorder, acute complications from substance use, homelessness, and HIV infection. Older patients and those with an overdose diagnosis had lower odds of receiving a consult. As shown in Table 1, the inverse probability of treatment weighted (IPTW) method using the propensity score improved balance compared with the unweighted sample for several covariates. (Table 1)

3.1. Primary analysis: 30-day acute care utilization

The repeat 30-day acute care utilization rate was 39.5% (274/694) among those encounters with an addiction consult and 36.0% (1902/5285) among those encounters without a consult (Unadjusted Odds Ratio 1.16; 95% CI 0.98-1.38). In adjusted analyses, an addiction consult was not statistically significantly associated with 30-day acute care utilization among SUD patient encounters (Adjusted Odds Ratio 1.02; 0.82, 1.28). (Table 2) We performed three sensitivity analyses to confirm the main findings, the first two included: i) using stabilized weights and ii) using stabilized weights truncated at a value of 10. A third sensitivity analysis explored the impact of variables included in the original propensity model whose temporality were unknown and may in fact have been a result of an addiction consult. Specifically we excluded the following variables from the propensity score: COWS, methadone during admission, buprenorphine during admission, and benzodiazepine or barbiturate during admission. The conclusions from these sensitivity analyses were consistent with the main analysis. (Appendix 1)

Table 2.

Unadjusted and adjusted association between Addiction Consult Service (ACS) and 30-day acute care utilization

Unadjusted OR (95% CI)* Adjusted OR (95% CI)*
Primary Analysis
Full Sample (N=3905 unique subjects, 5979 total observations) 1.16 (0.98, 1.38) 1.02 (0.82, 1.28)
Exploratory Subgroup Analyses
With OUD (N=1378 unique subjects, 2163 observations) 0.98 (0.79, 1.20) 0.87 (0.66, 1.13)
With OUD and medication (N=631 unique subjects, 952 observations) 0.83 (0.61, 1.12) 0.69 (0.47, 1.02)
With OUD without medication (N=850 unique subjects, 1211 observations) 1.16 (0.87, 1.55) 1.06 (0.73, 1.53)
Without OUD (N = 2654 unique subjects, 3816 observations) 1.04 (0.75, 1.44) 1.06 (0.74, 1.52)
Without OUD with medication (N = 249 unique subjects, 353 observations) 1.05 (0.64, 1.73) 1.09 (0.62, 1.90)
Without OUD without medication (N = 2476 unique subjects, 3463 observations) 0.84 (0.53, 1.34) 0.83 (0.50, 1.38)
Open in a new tab

Legend

*

All analyses performed reflect represent encounter-level analyses

OUD – Opioid Use Disorder

Medication – continued or started during admission, including the medications methadone, buprenorphine, naltrexone, acamprosate, disulfiram ortopiramate

Adjusted OR - represents an inverse probability of treatment weighting generalized estimating equation logistic model with weights truncated at 99th percentile, and includes the following variables in the propensity score: age, sex, race/ethnicity, insurance status, housing status, length of stay, ICU stay during admission, psychiatry consult during admission, discharge against medical advice, HIV status, types of substance use disordera, number of substance use disorders, naloxone administered, overdose diagnosis, acute complications, COWS ordered, outpatient methadone or buprenorphine, methadone or buprenorphine ordered during admissiona, benzodiazepine or barbiturate ordered during admission

a

Denotes variables removed from propensity score for secondary analysis of effect modification

3.2. Exploratory analyses: assessment of whether results differ by opioid use disorder (OUD) and medication for addiction treatment

Among 2163 admissions for patients with OUD, 952 (44.0%) received medication for addiction treatment on discharge. Overall, 3816 admissions for patients with a non-opioid use disorder occurred and of those 353 (9.3%) received medication for addiction treatment on discharge.

The 30-day acute care utilization rate for encounters with OUD was 41.7% (901/2163). The 30-day acute care utilization rate for encounters without OUD was 33.4% (1275/3816). Among the subgroup of patients with OUD, those who received a consult had a 13% decreased odds of any 30-day utilization (AOR 0.87; 0.66, 1.13) compared with those without an addiction consult, whereas among the subgroup of patients without OUD, those who received a consult had a 6% increased odds (AOR 1.06; 0.74, 1.52) compared with patients without an addiction consult. Neither result was statistically significant. (Table 2)

We then further stratified analyses by using the 4-category variable of opioid use disorder (OUD) and medication for addiction treatment. These generated the following adjusted odds ratios representing the association between receipt of ACS and 30-day acute care utilization: among patients with OUD and given medication, AOR 0.69 (CI 0.47, 1.02); among patients with OUD and without medication, AOR 1.06 (0.73, 1.53); among patients without OUD and given medication, AOR 1.09 (0.62, 1.90); and among patients without OUD and not given medication, AOR 0.83 (0.50, 1.38); within each of these 4 subgroups, the comparison of interest is receipt of an addiction consult compared with no addiction consult. (Table 2)

4. Discussion

Subsequent acute care utilization is common among hospitalized patients with SUD, particularly those who were referred to the ACS. This study found very high rate 30-day acute care utilization for patients with SUD overall: 39.5% for those with a consult and 36% for those not seen by the consult service. A similar study in a Maryland hospital found the 30-day readmission rate to be 13.9%.(Nordeck et al., 2018) That study did not include emergency department visits in its outcome, and emergency department utilization has been shown to be very high among patients with SUD.(Nambiar et al., 2018) Thus, the composite outcome used in this study - which includes both emergency department visits and re-hospitalization - may be a measure that is even more challenging to impact with an intervention such as an ACS. The patients also had substantial medical comorbidities with at least a quarter of admissions requiring intensive care, and over a third with significant medical complications of substance use. In an attempt to mitigate such confounding, both of these variables were controlled for in the propensity score analyses. In addition, as this was the first year of the ACS’s existence in this hospital, an addiction consult was a relatively uncommon event, with only one-fifth of primary SUD admissions involved an addiction service consult. This was new service and its early implementation may not have demonstrated the full potential of the ACS to efficiently and effectively deliver care and impact outcomes.

We hypothesized that the major impact of the ACS would be through the initiation of medications for addiction treatment, especially for patients with OUD. For the subset of patients with OUD who started or continued on medication, the odds of 30-day acute care utilization appeared to decrease by 31 % for those with receipt of ACS compared with those who did not. These findings, while not statistically significant, are in line with literature demonstrating that medications for OUD are associated with decreased hospitalizations and emergency department visits. However, prior studies demonstrating decreased care utilization among patients on medications for OUD included patients who engaged in outpatient buprenorphine treatment and remained in treatment for 12 months (Lo-Ciganic et al., 2016), or methadone maintenance patients in the first two years of treatment.(Russolillo et al., 2019) In this current study we grouped patients together who continued or initiated medication during the hospitalization because we were not able to distinguish them in the data. Presumably patients who newly initiated medications were more unstable in their recovery and overall than those continuing medications. Likely the vast majority of newly started medications were among patients seen by the ACS. Thus, grouping these patients may mitigate some of the impact of the ACS as ACS encounters for patients newly started medication were compared with encounters for patients who were more likely stable on medication and not seen by the ACS. Furthermore, we were not able to capture or explore the impact of duration of medication continuation after discharge. The medical complexity, social instability and limited treatment readiness of patients who are hospitalized and then offered addiction treatment makes this population quite distinct from patients who self-refer, engage and remain in long-term outpatient OUD treatment. However, a great strength of addiction consult teams is their ability to initiate medication for addiction treatment and link patients to long-term treatment. This increases the likelihood that hospitalized patients will reap the benefits of medications, which are associated with decreased morbidity and mortality.(Englander et al., 2019)

There are a myriad of pressures on hospitals to reduce readmissions.(Chokshi and Chang, 2014) Some hospital-based interventions have been shown to improve readmission rates,(Feltner et al., 2014; Gallagher et al., 2017) but others do not perform consistently. For example, a recent study of inpatient infectious disease consultation demonstrated decreased readmissions for some types of bacterial infections and not others.(Burnham et al., 2018) Because this is an observational study, consultants may be more likely to be called in for complex cases. This is similar to a study of heart failure patients, which found that a palliative care consultation increased 30-day readmission rates. The authors of that study attributed this to unmeasured confounders of disease severity and social determinants of health.(Chuang et al., 2017) There are likely some unmeasured confounders for which we were not able to fully account, given the data were collected retrospectively for clinical purposes. For example, social determinants of health including income, marital status and dependent children have large impact on readmission rates.(Meddings et al., 2017) Many of these variables were not available from the medical record in this retrospective study. Prior work has shown that OUD patients have higher rates of readmission. This may not be due to OUD alone, but could also be due to unmeasured confounders. (Nordeck et al., 2018)

The ACS was neither designed nor staffed to explicitly address acute care utilization. At the time of this study, the team consisted of only physicians assisted by one part-time nurse focusing on methadone linkage.(Trowbridge et al., 2017) The primary focus of the service was to provide harm reduction services and offer medications for SUD. More recent versions of the ACS, and other models around the country, include explicit use of peer recovery coaches, social workers and other staff. These additional team members are able to follow patients post-discharge and link patients to a wider array of outpatient services, which may be more likely to impact acute care utilization.(Collins et al., 2019; Weinstein et al., 2018)

The existence of the ACS itself may increase readmissions by creating a culture change in the hospital overall, making it a more welcoming and supportive place for patients with SUD, leading them to feel more comfortable returning, thus increasing readmissions. Notably, there was a high rate of AMA discharges among patients in this study (11.5% for those seen by ACS and 8.5% for those not seen). This is not surprising as AMAs are common among patients with SUD, especially those with OUD.(Lail and Fairbairn, 2018; McNeil et al., 2014; Ti and Ti, 2015; Zhu and Wu, 2019) In addition, while the ACS actively works to manage patients’ symptoms in the hospital and connect them to outpatient care, patients with SUD may also be at higher risk of readmission due to inability to receive appropriate medical care post-discharge. For example, many skilled nursing facilities are unwilling to accept patients with SUD.(Wakeman and Rich, 2017) Furthermore, concerns for a patient leaving AMA can be a reason for seeking an addiction service consult, which may contribute to this finding.

4.1. Strengths and Limitations

This study’s strengths included the use of propensity score analyses to minimize the confounding inherit in observational studies and to better estimate the heterogenous effect of an addiction consult. (D’Agostino, 1998) Another strength is the specific examination of the impact of providing medications in the hospital on post-discharge care utilization.

This study has some additional limitations. This includes our inability to capture acute care utilization at other hospitals, many of which are available in the metro Boston area.(Burke et al., 2018; Davies et al., 2013) Details about post-discharge addiction treatment engagement and retention were also not available for inclusion. Use of the propensity score analysis was a study strength; nonetheless, it may not have adequately controlled for confounding by all variables. Specifically, there may be residual confounding from variables such as whether benzodiazepines or barbiturates were ordered during admission, admitting service, type of SUD, and HIV status. For these variables, imbalance between groups remained even after propensity score adjustment (reflected in the standardized differences >.20 for these variables).(Ali et al., 2014; Austin and Stuart, 2015; Brooks and Ohsfeldt, 2013; D’Agostino, 1998) (Table 1) Also, as a consult service, the ACS does not submit any orders, but instead makes recommendations to the primary team. Some of the variables in the propensity score had an unknown temporality. For example, it is unknown whether the use of methadone in the hospital was in fact a recommendation by the ACS team or initiated prior to the consult by the primary team. However, the sensitivity analysis that removed the variables of unknown temporality was consistent with the main finding. In addition, there is still potential for unmeasured confounders, including confounding by indication related to the severity of SUD, and social determinants of health, which could impact readmission.(Brooks and Ohsfeldt, 2013) Although the overall sample size was large, this study may have been underpowered within the subgroup analyses to detect the impact of medications and type of SUD. A larger study over a longer period may be better able to capture the impact of an addiction consult service.

5. Conclusion

Repeat acute care utilization is common among hospitalized patients with SUD, including those who were referred to the addiction consult service. In adjusted analyses we did not detect a significant association between receipt of an addiction consult and 30-day acute care utilization. Our exploratory analyses suggest any impact of these services may be larger on patients with opioid use disorder who were started or continued on medications for addiction treatment. Addiction consult teams should recognize the possible impact of initiating medications for opioid use disorder and consider other specific interventions to decrease acute care utilization. Future research should include prospective studies designed to control for key confounders which are difficult to measure retrospectively, and should further examine the benefits of medications for substance use disorders.

Supplementary Material

1

Highlights:

  • Acute care utilization is common among patients with substance use disorders

  • No significant association between addiction consults and 30-day care utilization

  • Impact of medication for opioid use disorder on healthcare utilization merits further study

Acknowledgements:

The authors would like to thank Katherine Calver, PhD, for her assistance in editing the final manuscript.

Role of Funding Source: The project described was supported in part by the National Institute on Drug Abuse (Grant R25DA013582). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse. No funding sources had any role in any of the following: study design; the collection, analysis and interpretation of data; the writing of the report; or the decision to submit the paper for publication.

Footnotes

Conflict of Interest: No conflict declared.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Supplementary material can be found by accessing the online version of this paper at https://doi.org/10.1016/j.drugalcdep.2020.108081

References

  1. Agency for Healthcare Research and Quality, 2019. Statistical Brief #249. Healthcare Cost and Utilization Project (HCUP). [WWW Document]. URL www.hcup-us.ahrq.gov/reports/statbriefs/sb249-Mental-Substance-Use-Disorder-Hospital-Stays-2016.jsp?utm_source=ahrq&utm_medium=en1&utm_term=Stats&utm_content=1&utm_campaign=ahrq_en6_11_2019. (accessed 7.8.19). [PubMed] [Google Scholar]
  2. Ali MS, Groenwold RHH, Klungel OH, 2014. Propensity score methods and unobserved covariate imbalance: Comments on “Squeezing the balloon.” Health Serv. Res 49, 1074–1082. 10.1111/1475-6773.12152 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Appel PW, Ellison AA, Jansky HK, Oldak R, 2004. Barriers to enrollment in drug abuse treatment and suggestions for reducing them: opinions of drug injecting street outreach clients and other system stakeholders. Am. J. Drug Alcohol Abuse 30, 129–153. [DOI] [PubMed] [Google Scholar]
  4. Appel PW, Oldak R, 2007. A preliminary comparison of major kinds of obstacles to enrolling in substance abuse treatment (AOD) reported by injecting street outreach clients and other stakeholders. Am. J. Drug Alcohol Abuse 33, 699–705. 10.1080/00952990701522641 [DOI] [PubMed] [Google Scholar]
  5. Austin PC, Stuart EA, 2015. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat. Med 34, 3661–3679. 10.1002/sim.6607 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Baser O, Chalk M, Fiellin DA, Gastfriend DR, 2011. Cost and utilization outcomes of opioid-dependence treatments. Am. J. Manag. Care 17 Suppl 8, S235–248. [PubMed] [Google Scholar]
  7. Bell J, Turbow S, George M, Ali MK, 2017. Factors associated with high-utilization in a safety net setting. BMC Health Serv. Res 17, 273. 10.1186/s12913-017-2209-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brooks JM, Ohsfeldt RL, 2013. Squeezing the balloon: Propensity scores and unmeasured covariate ablance. Health Serv. Res 48, 1487–1507. 10.1111/1475-6773.12020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Burke RE, Jones CD, Hosokawa P, Glorioso TJ, Coleman EA, Ginde AA, 2018. Influence of nonindex hospital readmission on length of stay and mortality. Med. Care 56, 85–90. 10.1097/MLR.0000000000000829 [DOI] [PubMed] [Google Scholar]
  10. Burnham JP, Olsen MA, Stwalley D, Kwon JH, Babcock HM, Kollef MH, 2018. Infectious diseases consultation reduces 30-day and 1-year all-cause mortality for multidrug-resistant organism infections. Open Forum Infect. Dis 5. 10.1093/ofid/ofy026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Center for Health Information And Analysis, 2016. Behavioral Health & Readmissions in Massachusetts Acute Care Hospitals.
  12. Chokshi D, Chang J, 2014. Preventing early readmissions. JAMA 312, 1344–1345. 10.1001/jama.2014.9473 [DOI] [PubMed] [Google Scholar]
  13. Chuang E, Kim G, Blank AE, Southern W, Fausto J, 2017. 30-day readmission rates in patients admitted for heart failure exacerbation with and without palliative care consultation: A retrospective cohort study. J. Palliat. Med 20, 163–169. 10.1089/jpm.2016.0305 [DOI] [PubMed] [Google Scholar]
  14. Collins D, Alla J, Nicolaidis C, Gregg J, Gullickson DJ, Patten A, Englander H, 2019. “If It Wasn’t for Him, I Wouldn’t Have Talked to Them”: Qualitative Study of Addiction Peer Mentorship in the Hospital. J. Gen. Intern. Med 10.1007/s11606-019-05311-0 [DOI] [PubMed] [Google Scholar]
  15. D’Agostino RB, 1998. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat. Med 17, 2265–2281. [DOI] [PubMed] [Google Scholar]
  16. D’Amico MJ, Walley AY, Cheng DM, Forman LS, Regan D, Yurkovic A, Samet JH, Weinstein ZM, 2019. Which patients receive an addiction consult? A preliminary analysis of the INREACH (INpatient REadmission post-Addiction Consult Help) study. J. Subst. Abuse Treat 106, 35–42. 10.1016/j.jsat.2019.08.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Davies SM, Saynina O, McDonald KM, Baker LC, 2013. Limitations of using same-hospital readmission metrics. Int. J. Qual. Health Care J. Int. Soc. Qual. Health Care ISQua 25, 633–639. 10.1093/intqhc/mzt068 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Drainoni M-L, Farrell C, Sorensen-Alawad A, Palmisano JN, Chaisson C, Walley AY, 2014. Patient Perspectives of an Integrated Program of Medical Care and Substance Use Treatment. AIDS Patient Care STDs 28, 71–81. 10.1089/apc.2013.0179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Englander H, Dobbertin K, Lind BK, Nicolaidis C, Graven P, Dorfman C, Korthuis PT, 2019. Inpatient Addiction Medicine Consultation and Post-Hospital Substance Use Disorder Treatment Engagement: a Propensity-Matched Analysis. J. Gen. Intern. Med 10.1007/s11606-019-05251-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Feltner C, Jones CD, Cené CW, Zheng Z-J, Sueta CA, Coker-Schwimmer EJL, Arvanitis M, Lohr KN, Middleton JC, Jonas DE, 2014. Transitional care interventions to prevent readmissions for persons with heart failure: A systematic review and meta-analysis. Ann. Intern. Med 160, 774–784. 10.7326/M14-0083 [DOI] [PubMed] [Google Scholar]
  21. Gallagher NA, Fox D, Dawson C, Williams BC, 2017. Improving care transitions: complex high-utilizing patient experiences guide reform. Am. J. Manag. Care 23, e347–e352. [PubMed] [Google Scholar]
  22. Goldfield NI, McCullough EC, Hughes JS, Tang AM, Eastman B, Rawlins LK, Averill RF, 2008. Identifying potentially preventable readmissions. Health Care Financ. Rev 30, 75–91. [PMC free article] [PubMed] [Google Scholar]
  23. Gray ME, Rogawski McQuade ET, Scheld WM, Dillingham RA, 2018. Rising rates of injection drug use associated infective endocarditis in Virginia with missed opportunities for addiction treatment referral: a retrospective cohort study. BMC Infect. Dis 18. 10.1186/s12879-018-3408-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Gupta A, Nizamuddin J, Elmofty D, Nizamuddin SL, Tung A, Minhaj M, Mueller A, Apfelbaum J, Shahul S, 2018. Opioid abuse or dependence increases 30-day readmission rates after major operating room procedures: A national readmissions database study. Anesthesiol. J. Am. Soc. Anesthesiol 128, 880–890. 10.1097/ALN.0000000000002136 [DOI] [PubMed] [Google Scholar]
  25. Jones CM, Mack KA, Paulozzi LJ, 2013. Pharmaceutical overdose deaths, united states, 2010. JAMA 309, 657–659. 10.1001/jama.2013.272 [DOI] [PubMed] [Google Scholar]
  26. Lail P, Fairbairn N, 2018. Patients with substance use disorders leaving against medical advice: Strategies for improvement. J. Addict. Med 12, 421. 10.1097/ADM.0000000000000432 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lo-Ciganic W-H, Gellad WF, Gordon AJ, Cochran G, Zemaitis MA, Cathers T, Kelley D, Donohue JM, 2016. Association between trajectories of buprenorphine treatment and emergency department and in-patient utilization. Addiction 111, 892–902. 10.1111/add.13270 [DOI] [PubMed] [Google Scholar]
  28. Lynch FL, McCarty D, Mertens J, Perrin NA, Green CA, Parthasarathy S, Dickerson JF, Anderson BM, Pating D, 2014. Costs of care for persons with opioid dependence in commercial integrated health systems. Addict. Sci. Clin. Pract 9, 16. 10.1186/1940-0640-9-16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Marks LR, Munigala S, Warren DK, Liang SY, Schwarz ES, Durkin MJ, 2019. Addiction medicine consultations reduce readmission rates for patients with serious infections from opioid use disorder. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am 68, 1935–1937. 10.1093/cid/ciy924 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. McIntyre LK, Arbabi S, Robinson EF, Maier RV, 2016. Analysis of risk factors for patient readmission 30 days following discharge from general surgery. JAMA Surg. 151, 855–861. 10.1001/jamasurg.2016.1258 [DOI] [PubMed] [Google Scholar]
  31. McNeil R, Small W, Wood E, Kerr T, 2014. Hospitals as a ‘risk environment’: An ethno-epidemiological study of voluntary and involuntary discharge from hospital against medical advice among people who inject drugs. Soc. Sci. Med 105, 59–66. 10.1016/j.socscimed.2014.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Meddings J, Reichert H, Smith SN, Iwashyna TJ, Langa KM, Hofer TP, McMahon LF, 2017. The Impact of disability and social determinants of health on condition-specific readmissions beyond Medicare risk adjustments: A cohort study. J. Gen. Intern. Med 32, 71–80. 10.1007/s11606-016-3869-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Nambiar D, Spelman T, Stoové M, Dietze P, 2018. Are people who inject drugs frequent users of emergency department services? A cohort study (2008–2013). Subst. Use Misuse 53, 457–465. 10.1080/10826084.2017.1341921 [DOI] [PubMed] [Google Scholar]
  34. Nordeck CD, Welsh C, Schwartz RP, Mitchell SG, Cohen A, O’Grady KE, Gryczynski J, 2018. Rehospitalization and substance use disorder (SUD) treatment entry among patients seen by a hospital SUD consultation-liaison service. Drug Alcohol Depend. 186, 23–28. 10.1016/j.drugalcdep.2017.12.043 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Noska A, Mohan A, Wakeman S, Rich J, Boutwell A, 2015. Managing opioid use disorder during and after acute hospitalization: a case-based review clarifying methadone regulation for acute care settings. J. Addict. Behav. Ther. Rehabil 4. 10.4172/2324-9005.1000138 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Priest K, McCarty D, 2019. Role of the hospital in the 21st century opioid overdose epidemic: The addiction medicine consult service. J. Addict. Med 13, 104–112. 10.1097/ADM.0000000000000496 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Robins J, Hernán M, Brumback B, 2000. Marginal structural models and causal inference in epidemiology. Epidemiology 11, 550–560. [DOI] [PubMed] [Google Scholar]
  38. Robinson RL, Grabner M, Palli SR, Faries D, Stephenson JJ, 2016. Covariates of depression and high utilizers of healthcare: Impact on resource use and costs. J. Psychosom. Res 85, 35–43. 10.1016/j.jpsychores.2016.04.002 [DOI] [PubMed] [Google Scholar]
  39. Rudd RA, Aleshire N, Zibbell JE, Gladden RM, 2016. Increases in drug and opioid overdose deaths - United States, 2000-2014. MMWR Morb. Mortal. Wkly. Rep 64, 1378–1382. 10.15585/mmwr.mm6450a3 [DOI] [PubMed] [Google Scholar]
  40. Russolillo A, Moniruzzaman A, Somers JM, 2019. Association of methadone treatment with substance-related hospital admissions among a population in Canada with a history of criminal convictions. JAMA Netw. Open 2, e190595–e190595. 10.1001/jamanetworkopen.2019.0595 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Shanahan CW, Beers D, Alford DP, Brigandi E, Samet JH, 2010. A transitional opioid program to engage hospitalized drug users. J. Gen. Intern. Med 25, 803–808. 10.1007/s11606-010-1311-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Smith-Bernardin S, Rowe C, Behar E, Geier M, Washington S, Santos G-M, Euren J, Martin J, Gleghorn A, Coffin PO, 2018. Low-threshold extended-release naltrexone for high utilizers of public services with severe alcohol use disorder: A pilot study. J. Subst. Abuse Treat 85, 109–115. 10.1016/j.jsat.2017.11.005 [DOI] [PubMed] [Google Scholar]
  43. Stürmer T, Wyss R, Glynn RJ, Brookhart MA, 2014. JIM-13-0560 R1: Propensity scores for confounder adjustment when assessing the effects of medical interventions using non-experimental study designs. J. Intern. Med 275, 570–580. 10.1111/joim.12197 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Substance Abuse and Mental Health Services Administration, 2018. Key substance use and mental health indicators in the United States: Results from the 2017 National Survey on Drug Use and Health (HHS Publication No. SMA 18-5068, NSDUH Series H-53).
  45. The National Library of Medicine, 2019. Alcohol Disorder hOsPital Treatment Trial [WWW Document]. URL https://clinicaltrials.gov/ct2/show/NCT02478489 (accessed 7.11.19).
  46. Ti Lianping, Ti Lianlian, 2015. Leaving the Hospital Against Medical Advice Among People Who Use Illicit Drugs: A Systematic Review. Am. J. Public Health 105, e53–59. 10.2105/AJPH.2015.302885 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Trowbridge P, Weinstein ZM, Kerensky T, Roy P, Regan D, Samet JH, Walley AY, 2017. Addiction consultation services – Linking hospitalized patients to outpatient addiction treatment. J. Subst. Abuse Treat 79, 1–5. 10.1016/j.jsat.2017.05.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Volkow ND, Frieden TR, Hyde PS, Cha SS, 2014. Medication-assisted therapies — Tackling the opioid-overdose epidemic. N. Engl. J. Med 370, 2063–2066. 10.1056/NEJMp1402780 [DOI] [PubMed] [Google Scholar]
  49. Wakeman SE, Metlay JP, Chang Y, Herman GE, Rigotti NA, 2017. Inpatient addiction consultation for hospitalized patients increases post-discharge abstinence and reduces addiction severity. J. Gen. Intern. Med 1–8. 10.1007/s11606-017-4077-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wakeman SE, Rich JD, 2017. Barriers to post-acute care for patients on opioid agonist therapy: An example of systematic stigmatization of addiction. J. Gen. Intern. Med 32, 17–19. 10.1007/s11606-016-3799-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Walley AY, Alperen JK, Cheng DM, Botticelli M, Castro-Donlan C, Samet JH Alford DP, 2008. Office-based management of opioid dependence with buprenorphine: Clinical practices and barriers. J. Gen. Intern. Med 23, 1393–1398. 10.1007/s11606-008-0686-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Walley AY, Paasche-Orlow M, Lee EC, Forsythe S, Chetty VK, Mitchell S, Jack BW, 2012. Acute care hospital utilization among medical inpatients discharged with a substance use disorder diagnosis. J. Addict. Med 6, 50–56. 10.1097/ADM.0b013e318231de51 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wei J, Defries T, Lozada M, Young N, Huen W, Tulsky J, 2015. An inpatient treatment and discharge planning protocol for alcohol dependence: efficacy in reducing 30-Day readmissions and emergency department visits. J. Gen. Intern Med 30, 365–370. 10.1007/s11606-014-2968-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Weinstein ZM, Wakeman SE, Nolan S, 2018. Inpatient addiction consult service: Expertise for hospitalized patients with complex addiction problems. Med. Clin. North Am 102, 587–601. 10.1016/j.mcna.2018.03.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Wu L-T, Zhu H, Ghitza UE, 2018. Multicomorbidity of chronic diseases and substance use disorders and their association with hospitalization: Results from electronic health records data. Drug Alcohol Depend. 192, 316–323. 10.1016/j.drugalcdep.2018.08.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Yarborough BJH, Stumbo SP, McCarty D, Mertens J, Weisner C, Green CA, 2016. Methadone, buprenorphine and preferences for opioid agonist treatment: A qualitative analysis. Drug Alcohol Depend. 160, 112–118. 10.1016/j.drugalcdep.2015.12.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Zhu H, Wu L-T, 2019. Discharge against medical advice from hospitalizations for substance use disorders: The potential impact of the Affordable Care Act. Drug Alcohol Depend. 197, 115–119. 10.1016/j.drugalcdep.2018.12.032 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1599421-supplement-1.docx (14.2KB, docx)

RESOURCES